XML: Parsing and Writing

5.3.6

top ← prev up next →

XML: Parsing and Writing

Paul Graunke and Jay McCarthy

The xml library provides functions for parsing and generating XML. XML can be represented as an instance of the document structure type, or as a kind of S-expression that is called an X-expression.

The xml library does not provide Document Type Declaration (DTD) processing, including preservation of DTDs in read documents, or validation. It also does not expand user-defined entities or read user-defined entities in attributes. It does not interpret namespaces either.

1 Datatypes

struct
(struct location (line char offset)
#:extra-constructor-name make-location)
  line : (or/c false/c exact-nonnegative-integer?)
  char : (or/c false/c exact-nonnegative-integer?)
  offset : exact-nonnegative-integer?

Represents a location in an input stream. The offset is a character offset unless xml-count-bytes is #t, in which case it is a byte offset.

value
location/c : contract?

Equivalent to (or/c location? symbol? false/c).

struct
(struct source (start stop)
#:extra-constructor-name make-source)
start : location/c
stop : location/c

Represents a source location. Other structure types extend source.

When XML is generated from an input stream by read-xml, locations are represented by location instances. When XML structures are generated by xexpr->xml, then locations are symbols.

struct
(struct external-dtd (system)
#:extra-constructor-name make-external-dtd)
system : string?
struct
(struct external-dtd/public external-dtd (public)
#:extra-constructor-name make-external-dtd/public)
public : string?
struct
(struct external-dtd/system external-dtd ()
#:extra-constructor-name make-external-dtd/system)

Represents an externally defined DTD.

struct
(struct document-type (name external inlined)
#:extra-constructor-name make-document-type)
  name : symbol?
  external : external-dtd?
  inlined : false/c

Represents a document type.

struct
(struct comment (text)
#:extra-constructor-name make-comment)
text : string?

Represents a comment.

struct
(struct p-i source (target-name instruction)
#:extra-constructor-name make-p-i)
target-name : symbol?
instruction : string?

Represents a processing instruction.

value
misc/c : contract?

Equivalent to (or/c comment? p-i?)

struct
(struct prolog (misc dtd misc2)
#:extra-constructor-name make-prolog)
  misc : (listof misc/c)
  dtd : (or/c document-type false/c)
  misc2 : (listof misc/c)

Represents a document prolog.

struct
(struct document (prolog element misc)
#:extra-constructor-name make-document)
  prolog : prolog?
  element : element?
  misc : (listof misc/c)

Represents a document.

struct
(struct element source (name attributes content)
#:extra-constructor-name make-element)
  name : symbol?
  attributes : (listof attribute?)
  content : (listof content/c)

Represents an element.

struct
(struct attribute source (name value)
#:extra-constructor-name make-attribute)
name : symbol?
value : (or/c string? permissive/c)

Represents an attribute within an element.

value
content/c : contract?

Equivalent to (or/c pcdata? element? entity? comment? cdata? p-i? permissive/c).

value
permissive/c : contract?

If (permissive-xexprs) is #t, then equivalent to any/c, otherwise equivalent to (make-none/c 'permissive)

procedure
(valid-char? x) → boolean?
x : any/c

struct
(struct entity source (text)
#:extra-constructor-name make-entity)
text : (or/c symbol? valid-char?)

Represents a symbolic or numerical entity.

struct
(struct pcdata source (string)
#:extra-constructor-name make-pcdata)
string : string?

Represents PCDATA content.

struct
(struct cdata source (string)
#:extra-constructor-name make-cdata)
string : string?

Represents CDATA content.

The string field is assumed to be of the form <![CDATA[‹content›]]> with proper quoting of ‹content›. Otherwise, write-xml generates incorrect output.

struct
(struct exn:invalid-xexpr exn:fail (code)
#:extra-constructor-name make-exn:invalid-xexpr)
code : any/c

Raised by validate-xexpr when passed an invalid X-expression. The code fields contains an invalid part of the input to validate-xexpr.

struct
(struct exn:xml exn:fail:read ()
#:extra-constructor-name make-exn:xml)

Raised by read-xml when an error in the XML input is found.

procedure
(xexpr? v) → boolean?
v : any/c

Returns #t if v is a X-expression, #f otherwise.

The following grammar describes expressions that create X-expressions:

xexpr	=	string
	\|	(list symbol (list (list symbol string) ...) xexpr ...)
	\|	(cons symbol (list xexpr ...))
	\|	symbol
	\|	valid-char?
	\|	cdata
	\|	misc

A string is literal data. When converted to an XML stream, the characters of the data will be escaped as necessary.

A pair represents an element, optionally with attributes. Each attribute’s name is represented by a symbol, and its value is represented by a string.

A symbol represents a symbolic entity. For example, 'nbsp represents  .

An valid-char? represents a numeric entity. For example, #x20 represents .

A cdata is an instance of the cdata structure type, and a misc is an instance of the comment or p-i structure types.

value
xexpr/c : contract?

A contract that is like xexpr? except produces a better error message when the value is not an X-expression.

2 X-expression Predicate and Contract

(require xml/xexpr)

The xml/xexpr library provides just xexpr/c, xexpr?, correct-xexpr?, and validate-xexpr from xml with minimal dependencies.

3 Reading and Writing XML

procedure
(read-xml [in]) → document?
in : input-port? = (current-input-port)

Reads in an XML document from the given or current input port XML documents contain exactly one element, raising xml-read:error if the input stream has zero elements or more than one element.

Malformed xml is reported with source locations in the form ‹l›.‹c›/‹o›, where ‹l›, ‹c›, and ‹o› are the line number, column number, and next port position, respectively as returned by port-next-location.

Any non-characters other than eof read from the input-port appear in the document content. Such special values may appear only where XML content may. See make-input-port for information about creating ports that return non-character values.

Example:

> (xml->xexpr (document-element
(read-xml (open-input-string
"<doc><bold>hi</bold> there!</doc>"))))
'(doc () (bold () "hi") " there!")

procedure
(read-xml/document [in]) → document?
in : input-port? = (current-input-port)

Like read-xml, except that the reader stops after the single element, rather than attempting to read "miscellaneous" XML content after the element. The document returned by read-xml/document always has an empty document-misc.

procedure
(read-xml/element [in]) → element?
in : input-port? = (current-input-port)

Reads a single XML element from the port. The next non-whitespace character read must start an XML element, but the input port can contain other data after the element.

procedure
(syntax:read-xml [in]) → syntax?
in : input-port? = (current-input-port)

Reads in an XML document and produces a syntax object version (like read-syntax) of an X-expression.

procedure
(syntax:read-xml/element [in]) → syntax?
in : input-port? = (current-input-port)

Like syntax:real-xml, but it reads an XML element like read-xml/element.

procedure
(write-xml doc [out]) → void?
doc : document?
out : output-port? = (current-output-port)

Writes a document to the given output port, currently ignoring everything except the document’s root element.

procedure
(write-xml/content content [out]) → void?
content : content/c
out : output-port? = (current-output-port)

Writes document content to the given output port.

procedure
(display-xml doc [out]) → void?
doc : document?
out : output-port? = (current-output-port)

Like write-xml, but newlines and indentation make the output more readable, though less technically correct when whitespace is significant.

procedure
(display-xml/content content [out]) → void?
content : content/c
out : output-port? = (current-output-port)

Like write-xml/content, but with indentation and newlines like display-xml.

procedure
(write-xexpr xe [out]) → void?
xe : xexpr/c
out : output-port? = (current-output-port)

Writes an X-expression to the given output port, without using an intermediate XML document.

4 XML and X-expression Conversions

parameter
(permissive-xexprs) → boolean?
(permissive-xexprs v) → void?
v : any/c

If this is set to non-false, then xml->xexpr will allow non-XML objects, such as other structs, in the content of the converted XML and leave them in place in the resulting “X-expression”.

procedure
(xml->xexpr content) → xexpr/c
content : content/c

Converts document content into an X-expression, using permissive-xexprs to determine if foreign objects are allowed.

procedure
(xexpr->xml xexpr) → content/c
xexpr : xexpr/c

Converts an X-expression into XML content.

procedure
(xexpr->string xexpr) → string?
xexpr : xexpr/c

Converts an X-expression into a string containing XML.

procedure
(string->xexpr str) → xexpr/c
str : string?

Converts XML represented with a string into an X-expression.

procedure
((eliminate-whitespace [tags choose]) elem) → element?
  tags : (listof symbol?) = empty
  choose : (boolean? . -> . boolean?) = (λ (x) x)
  elem : element?

Some elements should not contain any text, only other tags, except they often contain whitespace for formating purposes. Given a list of tag names as tags and the identity function as choose, eliminate-whitespace produces a function that filters out PCDATA consisting solely of whitespace from those elements, and it raises an error if any non-whitespace text appears. Passing in not as choose filters all elements which are not named in the tags list. Using (lambda (x) #t) as choose filters all elements regardless of the tags list.

procedure
(validate-xexpr v) → #t
v : any/c

If v is an X-expression, the result is #t. Otherwise, exn:invalid-xexprs is raised, with a message of the form “Expected ‹something›, given ‹something-else›”. The code field of the exception is the part of v that caused the exception.

Examples:

> (validate-xexpr '(doc () "over " (em () "9000") "!"))
#t
> (validate-xexpr #\newline)
Expected a string, symbol, valid numeric entity, comment,
processing instruction, or list, given #\newline

procedure
(correct-xexpr? v success-k fail-k) → any/c
  v : any/c
  success-k : (-> any/c)
  fail-k : (exn:invalid-xexpr? . -> . any/c)

Like validate-xexpr, except that success-k is called on each valid leaf, and fail-k is called on invalid leaves; the fail-k may return a value instead of raising an exception or otherwise escaping. Results from the leaves are combined with and to arrive at the final result.

5 Parameters

parameter
(empty-tag-shorthand)
→ (or/c (one-of/c 'always 'never) (listof symbol?))
(empty-tag-shorthand shorthand) → void?
shorthand : (or/c (one-of/c 'always 'never) (listof symbol?))

A parameter that determines whether output functions should use the <‹tag›/> tag notation instead of <‹tag›></‹tag›> for elements that have no content.

When the parameter is set to 'always, the abbreviated notation is always used. When set of 'never, the abbreviated notation is never generated. when set to a list of symbols is provided, tags with names in the list are abbreviated.

The abbreviated form is the preferred XML notation. However, most browsers designed for HTML will only properly render XHTML if the document uses a mixture of the two formats. The html-empty-tags constant contains the W3 consortium’s recommended list of XHTML tags that should use the shorthand. This list is the default value of empty-tag-shorthand.

value
html-empty-tags : (listof symbol?)

See empty-tag-shorthand.

Example:

> (parameterize ([empty-tag-shorthand html-empty-tags])
    (write-xml/content (xexpr->xml `(html
                                      (body ((bgcolor "red"))
                                        "Hi!" (br) "Bye!")))))
<html><body bgcolor="red">Hi!<br />Bye!</body></html>

parameter
(collapse-whitespace) → boolean?
(collapse-whitespace collapse?) → void?
collapse? : any/c

A parameter that controls whether consecutive whitespace is replaced by a single space. CDATA sections are not affected. The default is #f.

parameter
(read-comments) → boolean?
(read-comments preserve?) → void?
preserve? : any/c

A parameter that determines whether comments are preserved or discarded when reading XML. The default is #f, which discards comments.

parameter
(xml-count-bytes) → boolean?
(xml-count-bytes count-bytes?) → void?
count-bytes? : any/c

A parameter that determines whether read-xml counts characters or bytes in its location tracking. The default is #f, which counts characters.

You may want to use #t if, for example, you will be communicating these offsets to a C program that can more easily deal with byte offsets into the character stream, as opposed to UTF-8 character offsets.

parameter
(xexpr-drop-empty-attributes) → boolean?
(xexpr-drop-empty-attributes drop?) → void?
drop? : any/c

Controls whether xml->xexpr drops or preserves attribute sections for an element that has no attributes. The default is #f, which means that all generated X-expression elements have an attributes list (even if it’s empty).

6 PList Library

(require xml/plist)

The xml/plist library provides the ability to read and write XML documents that conform to the plist DTD, which is used to store dictionaries of string–value associations. This format is used by Mac OS X (both the operating system and its applications) to store all kinds of data.

A plist value is a value that could be created by an expression matching the following pl-expr grammar, where a value created by a dict-expr is a plist dictionary:

pl-expr	=	string
	\|	(list 'true)
	\|	(list 'false)
	\|	(list 'integer integer)
	\|	(list 'real real)
	\|	dict-expr
	\|	(list 'array pl-expr ...)

dict-expr	=	(list 'dict assoc-pair ...)

assoc-pair	=	(list 'assoc-pair string pl-expr)

procedure
(plist-value? any/c) → boolean?
any/c : v

Returns #t if v is a plist value, #f otherwise.

procedure
(plist-dict? any/c) → boolean?
any/c : v

Returns #t if v is a plist dictionary, #f otherwise.

procedure
(read-plist in) → plist-value?
in : input-port?

Reads a plist from a port, and produces a plist value.

procedure
(write-plist dict out) → void?
dict : plist-value?
out : output-port?

Write a plist value to the given port.

Examples:

> (define my-dict
    `(dict (assoc-pair "first-key"
                       "just a string with some  whitespace")
           (assoc-pair "second-key"
                       (false))
           (assoc-pair "third-key"
                       (dict))
           (assoc-pair "fourth-key"
                       (dict (assoc-pair "inner-key"
                                         (real 3.432))))
           (assoc-pair "fifth-key"
                       (array (integer 14)
                              "another string"
                              (true)))
           (assoc-pair "sixth-key"
                       (array))))
> (define-values (in out) (make-pipe))
> (write-plist my-dict out)
> (close-output-port out)
> (define new-dict (read-plist in))
> (equal? my-dict new-dict)
#t

The XML generated by write-plist in the above example looks like the following, if re-formatted by:

<?xml version="1.0" encoding="UTF-8"?>

<!DOCTYPE plist SYSTEM

"file://localhost/System/Library/DTDs/PropertyList.dtd">

<dict>

<key>first-key</key>

<string>just a string with some whitespace</string>

<key>second-key</key>

<key>third-key</key>

<key>fourth-key</key>

<dict>

<key>inner-key</key>

</dict>

<key>fifth-key</key>

<array>

<string>another string</string>

</array>

<key>sixth-key</key>

</dict>

</plist>

7 Simple X-expression Path Queries

(require xml/path)

This library provides a simple path query library for X-expressions.

value
se-path? : contract?

A sequence of symbols followed by an optional keyword.

The prefix of symbols specifies a path of tags from the leaves with an implicit any sequence to the root. The final, optional keyword specifies an attribute.

procedure
(se-path*/list p xe) → (listof any/c)
p : se-path?
xe : xexpr?

Returns a list of all values specified by the path p in the X-expression xe.

procedure
(se-path* p xe) → any/c
p : se-path?
xe : xexpr?

Returns the first answer from (se-path*/list p xe).

Examples:

> (define some-page
    '(html (body (p ([class "awesome"]) "Hey") (p "Bar"))))
> (se-path*/list '(p) some-page)
'("Hey" "Bar")
> (se-path* '(p) some-page)
"Hey"
> (se-path* '(p #:class) some-page)
"awesome"
> (se-path*/list '(body) some-page)
'((p ((class "awesome")) "Hey") (p "Bar"))
> (se-path*/list '() some-page)
'((html (body (p ((class "awesome")) "Hey") (p "Bar")))
  (body (p ((class "awesome")) "Hey") (p "Bar"))
  (p ((class "awesome")) "Hey")
  "Hey"
  (p "Bar")
  "Bar")

top ← prev up next →

1	Datatypes
2	X-expression Predicate and Contract
3	Reading and Writing XML
4	XML and X-expression Conversions
5	Parameters
6	PList Library
7	Simple X-expression Path Queries